[R-sig-Geo] Internal limits of processing in geoR ?

[Roger Bivand]

On Wed, 6 Jun 2007, Vanessa Stelzenmuller (Cefas) wrote:

> Hello everyone,
> 
> I'm new to this list and I wonder if someone could help with the following:
> 
> 
> - I have to process a huge amount of data(> 20 000) - I calculated some
> GAMs, correlations etc. no problem
> 
> - Now using the library (geoR) I came across the following error message
> for the first time: "cannot allocate vector of size 567 MB"
> 
> - It is not a problem of computer power - so I guess these are internal
> limits for processing with geoR?
> 
> What do you think ?
> 

RSiteSearch("geoR memory")

gives some helpful links. The functions in geoR do work on the dense n by 
n matrix, so may not be what you need - depending on the particular 
geostatistical approach you are taking. If you are free to use 
alternatives. please look at gstat (and I believe fields) which do not try 
to store the n by n matrix. 

geoR is, I believe, best in situations where there is little data and the
researcher is trying to squeeze the maximum out of them in terms of
inference and prediction - see the geoR book "Model-based geostatistics"  
by Peter Diggle and Paulo Ribeiro Jr, recently published by Springer.

Roger

> 
>  
> 
> <º)))))>< <º))))>< <º)))>< <º))><
> 
> 
>    Dr. Vanessa Stelzenmüller
>    CEFAS, Pakefield Road
>    Lowstoft, NR33 OHT, UK
> 
>    phone: + 44 (0)1502 527779
>    fax:     + 44 (0)1502 513865
> 
> <º))>< <º)))>< <º))))>< <º)))))>< 
> 
> 
> ***********************************************************************************
> This email and any attachments are intended for the named re...{{dropped}}
> 
> 

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no